Introduction There is leverage to be gained by bringing spatial data within the purview of general purpose database systems. A spatial access method embedded in a general purpose DBMS would have several advantages [Lom91]: 1. Spatial data could be integrated with other data. Spatial objects are likely to have attributes other than the location of their bounding boxes. These other attributes can be queried using traditional methods. 2. Multiple users (some making updates) would be supported by concurrency algorithms already in place. 3. In case of a system failure, the restart process would be able to recover the database to a consistent state. 4. Robust system utilities for loading data, analyzing performance, producing reports and so forth could be applied. To this end, we have been modifying the hB-tree (or holey Brick tree) [LS90], an efficient multiattribute search structure, for use with the concurrency and recovery systems available in general purpose DBMSs. This wi